The Logic of Anarchy: Neorealism to Structural Realism

Barry Buzan, Charles Jones and Richard Little

New York

Columbia University Press

1993

10. Analogy, Theory, and Testing

 

A majority of social scientists might accept that within the frontiers of the nation state there is a great variety of sources and expressions of social power. Yet many, even of those who hold this position with regard to domestic politics, still accept a Realist view of international relations, regarding coercive force exerted by functionally homogeneous states as ultimately decisive within an anarchic system.

 

Realism: Discourse and Policy

The breathtakingly audacious appropriation by the modern state of processes of enforcement and of concepts of law, regulation, and order to which this general consensus bears witness has made out of International Relations possibly the last redoubt of unregenerate structuralism in the social sciences. This appropriation has been overwhelmingly successful in modern history. Resistance to it has generally taken the form of attempts, within the territory of particular states, to salvage some distinctiveness for the concepts--and realities--of civil society and personal emancipation in the face of the omnipresent totalizing state. This is probably why relations between states have often been marginalized, their study too frequently deferred, so that they have been able even at this late date to seem simple, and therefore ideally susceptible to heroic abstraction.

Moreover this state-centricity has identified the Realist approach especially closely in recent years with a politically conservative strand of structuralism and with the positivistic methodology thought bound to accompany it. Realism is ideology to the extent that its employment of the mental trick of abstraction "obscures the real condition of society . . . and thereby stabilizes it" (Mannheim 1936:36). But Realism need not be ideology. It may even, if those familiar with the terminology of International Relations will forgive what must at first seem a paradox, be utopian; better, since utopianism is too often little more than ideology wearing a smile, it may instead be emancipatory.

One reason that Kenneth Waltz's TIP led to such prolonged and often tetchy debate during the 1980s was that he inadvertently opened up a battle front between those within the coercive-state consensus, who thought a severe structuralist approach, once clearly enunciated, to be so obvious as to require no defense, and those, often from elsewhere in the social sciences, who had believed structuralism to be already brain-dead, only waiting for its Cold War life-support machine to be turned off.

One intention of this book has been to move beyond the blank mutual incomprehension and evident distaste which were present in some of the early exchanges. In Sections I and II an attempt has already been made to entertain the flux and uncertainty of events, to avoid privileging social structures over the actors who constitute, reproduce, and transform them, and to differentiate, even within the international anarchy, between various forms of power based in a multiplicity of social formations, including the sectoral divisions of the international system offered in Section I. We have contended that all this sums to a reintegration of serious concern with social structure--albeit now in a more subtle and guarded form than Mark-1 structuralism--into a Realist tradition that has always had at its core the unity of deliberation about society and social action: theory and practice.

Because of this essentially conciliatory purpose, and the gentle tone to which it gives rise, the disagreements about many facets of a Neorealist theory of international politics which will so clearly distinguish the authors of this volume from many of their recent Realist predecessors in the eyes of those familiar with the literature may easily appear to the more casual reader as mere family disagreements or scholarly nuances. Yet any such interpretation would miss a most fundamental difference of emphasis concerning the logical status claimed here for any Structural Realist analysis or theory. That is why this Section of the book tries, starting from an examination of the epistemological and methodological assumptions underlying the analogy with microeconomics employed in Waltz's TIP, to argue that any theory of international politics is much less a passive representation of unalterable conditions of the social world it treats than a means to the active shaping of that world.

In E. H. Carr's terms, power was divisible into political power, economic power, and power over opinion. In our terms power, which cannot always be reduced to state power, is manifested in several discrete though related sectors of the international system, including the cultural or ideological sector. Though it will often be ineffectual, discourse about international relations amounts to endogenous action within the policy-making process; it is part of the game, never the innocent exposure by disinterested Martians of exogenous constraints upon human policy. In short, what is proposed here is a reaffirmation of the pragmatism and of the unity of purpose and analysis which, even more than its emphasis on the state, constitute the core of political Realism (Carr 1939:6).

All this needs spelling out a little. Carr long ago made clear that the price of exposing the socially constituted character of other people's perceptions of international politics was an admission that one's own views, too, arose within a prior context of culture and interest. Knowledge of society was always relative to one's position in society. Great powers produced ingenious and self-serving arguments to justify the status quo; revisionist powers produced equally ingenious and self-serving critiques of the dominant ideology. Reasoning along these lines had led Karl Mannheim, on whom Carr was to rely so heavily, to exclaim that "political discussion is, from the very first . . . the tearing off of disguises" (Mannheim 1936:35).

Empiricists might be inclined to make light of this. If there is an independent way of testing knowledge claims against the world then it makes sense to reply to Carr that why people hold the views they do is a quite different and perhaps a less important question than whether or not those views are true. But the statesman who, perhaps unconsciously, adopts a rationalist or pragmatist view of knowledge will not escape so easily. Driven, like Carr, to the conclusion that Realism is "just as much conditioned as any other mode of thought," he may feel unable to make universal claims and pursue clear policies based upon them, and so fall prey to less sophisticated opponents who lack such scruples, remaining fully and uncritically convinced of the justice and enduring truth of their claims (Carr 1939:113). It is precisely this sort of moral funk, widely but wrongly believed to follow remorselessly from acceptance of a social constructivist approach to knowledge, which has led many to regard any move along that path as an assault on reason itself.

Alternatively, however, one may take the much more positive view of the relativist dilemma outlined by Mannheim in a work Carr knew well, finding empowerment in the recognition of constraint (Mannheim 1936:42-43). His argument is that discovering the manner and extent of the social determination of one's thought and behavior is a liberating experience--almost a secular revelation--because it brings within reach the explicit choice of struggling to surmount that determination by transforming the social structures through which it operates. Think of knowledge as a social structure and the individual knower as agent, and it will be apparent that this is very much the picture we have been trying to develop in the treatment of structure and agency offered in Section II. All this sounds grand (or vacuous) until one realizes that it is the kind of thing people do all the time, walking across grass, and so creating a new path, because they realize that the old one, however convenient for their predecessors, is taking them out of their way. And calling them "deep" or "generative" renders the major continuities of international politics by which Waltz is almost bewitched no less specifically historical, no less optional, and no less dependent on our behavior for their continued reproduction.

It is impossible to describe the process of political deliberation without at the same time contributing to it in some measure. The idea that there is some more objective meta-discussion of political discourse which is itself removed from the struggle is precisely the illusion that Carr and Mannheim strove to expose. Yet often the political effectiveness of such descriptions appears to depend precisely on their being perceived as something quite apart from the policy-making process, with an authority distinct from and higher than that of world leaders and diplomats smirched by close engagement in political action: an authority deriving from their status as History or Theory or Science.

This argument may be flawed; after all, understanding how the advertisement manipulates need not necessarily stop one from choosing freely to buy the product all the same. But, flawed or clear, it has been widely accepted in practice. Thus Carr conceals his plea for the admission of the Soviet Union to the Great Power club behind what was, to his target audience of British Conservative politicians, a more acceptable, though precisely parallel, argument for appeasing Hitler's Germany. He then grounds both in an overarching historicism. In the meantime, Carr's explicit treatment of propaganda is, to say the least, ingenuous and underdeveloped when judged by his own employment of rhetorical technique.

Waltz, for his part, cast his topical appeal for continuity and responsibility in United States foreign policy in a mold of enduring and supposedly universal truths. In each case, history or theory seems to be politically effective precisely because and to the extent that its instrumentality is concealed beneath a wash of science. There is a persistent tension between the requirements and standards of analysis and prescription, science and policy, understanding and action, the academic and the politician.

The argument that follows will therefore be that the Realist tradition of analysis of international relations consists, and is bound to consist, not simply in radical un-masking of the kind suggested by Mannheim, nor in "the conservative ideology of the exercise of state power" with which it was recently identified by Justin Rosenberg, but in the taking off and putting on of masks by an alternation of creation and criticism, exposure and concealment, conjecture and formalism, utopianism and ideology (Rosenberg 1990:296). Kenneth Waltz's Neorealist theory of international politics, like Carr's Twenty Years' Crisis, came at a moment when circumstances were judged by the author to require criticism, concealment, and formalism. This seemed during the brief months between the fall of the Berlin Wall and the invasion of Kuwait to be less the case than at any time since 1914. Nevertheless, even the most masked and statist of Realists, Carr and Waltz included, have always provided possibly unintended clues to escape routes by the use of metaphors that put the state in question by identifying it as person, as actor, or as firm; and it is through an examination of these metaphors, the last most of all, that an attempt will be made to complete and ground our liberalizing move from Neorealism to Structural Realism.

 

The Analogy with Economics

Throughout TIP Waltz makes extensive use of an analogy or formal comparison between balance of power theory and microeconomics (Waltz 1979, 54-55, 72-74, 89-94 and passim). Microeconomics is the study of interactions between economic agents in markets. It is generally contrasted with macroeconomics, the study of the performance of the whole (national!) economy. Economists still start their students off in microeconomics with the model of a free and perfect market as a frame of reference: "free" because there is no bar to entry, "perfect" because no single agent can have a perceptible influence on prices and all must therefore respond to the sovereign market as price-takers. No one pretends that such a market exists, and the student is accordingly led step by step into the more complex and realistic world of imperfect competition. Here--where the number of firms in the market may be a few (oligopoly), just two (duopoly), or one alone (monopoly)--possibilities of manipulation, collusion, and strategy abound. Each actor must be constantly aware of others' actions. Each must calculate the variety of responses to be anticipated in reaction to its own moves. The game has seemed to many observers, Waltz included, to bear an uncanny resemblance to the maneuverings of the balance of power. Oligopoly, with a handful of firms accounting for sixty or seventy percent of the market or more, is seen to resemble the pre-1939 balance; duopoly, with just two firms dominating the market, is readily compared with bipolarity.

Argument by analogy has long held a prominent place in writings about international relations; this, in addition to its centrality in Waltz's TIP, is why it has been selected for close examination here. Most familiar is the domestic analogy, where states in an international anarchy are compared to individual persons in a state of nature. More recently, historical analogies have been popular, and students have been invited to examine alliance systems or to compare the pax Britannica and the pax Americana, asking whether each depended upon the provision of international public goods by a hegemonic power (Kindleberger 1973; Olson and Zeckhauser 1966). Yet one seldom reads a clear account in the International Relations literature of why it is that structural similarity in one sphere of human activity should offer insights into other activities which are similarly structured, or of how one is to decide just which features of a structure are to be transposed in the analogy and which are trivial or irrelevant. Equally, when arguments by analogy are attacked, the attack is usually piecemeal rather than systematic.

Waltz is distinctive, therefore, not so much in his use of analogy as in his justification of it. He provides an explicit account of the methodological basis for argument by analogy, and our next task is to examine that account. It will be argued that Waltz's attempt to use analogy as a method of confirmation within a positivist understanding of scientific theory is incoherent, and indeed that his whole treatment of the testing of theory, both by analogy and by hypothesis, is inconsistent with the greater and better part of his own methodological recipe. There is a ground-clearing job to be done.

 

Some Traditional Categories

"Reasoning by analogy is helpful," Waltz writes, "where one can move from a domain for which theory is well developed to one where it is not. Reasoning by analogy is permissible where different domains are structurally similar" (Waltz, 1979:89). By his own rules, then, the analogy between balance of power theory and microeconomics that pervades his work requires Waltz to satisfy the reader on several criteria. What is a well-developed theory and how is it recognized? How is one domain distinguished from another? What is it for two domains to be structurally similar? Does the analogy between balance of power systems and markets satisfy these criteria of similarity? And just what kind of help does a legitimate analogy provide? Is its function that of theory? Does it explain? Does it suggest hypotheses? Does it confirm? Or what? Again, how important anyway is this argument to Waltz's theory? Are his central contentions threatened if this bastion falls?

The first of these questions plunges us straight away into something of a mire. The best way to begin would appear to be to establish what would count for Waltz as a well-developed theory, and to see what his views on this subject commit him to. The epistemological question is fundamental here, because methodology, best defined as procedures of investigation and standards of explanation, generally follows from epistemology. How you set about establishing truths is a function of what you think knowledge consists in. But the account both of knowledge and theory which Waltz provides is a hybrid, uneasily combining at least two incompatible epistemological traditions. Because of this, and because his positivist attitudes toward testing (methodology) might automatically though wrongly brand him for many readers as an empiricist (epistemology), a few paragraphs must first be devoted to clarification of the terms in which the analysis of these questions has customarily been couched. Three are used here--"empiricist," "rationalist," and "pragmatist"--to refer to broad attitudes toward the nature of human knowledge, or to use the philosophers' neater term, epistemologies. To avoid confusion it should at once be said that for much of the modern period the problem of knowledge and the quest for secure foundations of knowledge--for certainty--has been central to Western philosophy and that each of these positions stems from this core epistemological project. Yet now it no longer holds the center of the stage. Indeed, it will be argued below that ontological, rather than epistemological, distinctions provide the ground of social inquiry and account for its peculiar richness and difficulty. More of this later.

Empiricists take the view that knowledge derives from experience. We receive information through our senses, and whatever we know of the world is ultimately derived from these sense data. The central function of language is simply to represent experience, and it is possible, in principle, for language to function as a transparent medium facilitating a near-complete and wholly literal description of the world. Our knowledge is regarded as a representation--doubtless sometimes erroneous or incomplete--but a representation none the less, corresponding to the external world.

(Notice already how shot through with metaphor is this description. We use the phrase "external world" without thinking about it. Is the perceiving self material? [a series of brain-states, perhaps?] If so, it, too, is part of the world that can only be known about by experience: that is, the "external" world. Outside what? Is it instead something incorporeal? Then it has no location, and the spatial metaphor of internal/external must be taken as figurative. There are not two things here, each of the same kind, one of which is inside and one outside. The figure [internal/external] is a way of covering up or smoothing over an incoherence in a great deal of our talk about perception and knowledge. Exposing this is not equivalent to peeling off a rotten layer of the onion to reveal clean white growth beneath, nor need it make any difference to our beliefs, behavior, or talk, for there is no obvious literal clarity beneath the metaphor [Beneath?!].)

The rationalist, while still intent on the quarry of certainty, gives a much more prominent role in the creation of knowledge to the knower. Rationalists very often set out from the question: "How can anyone make sense of the constant stream of sense data?" The argument is that there are certain categories--space, time, number, and the like--which must logically be prior to knowledge, since without them we could make no sense of our experience. One way of putting this is to say that though there may well be a world "out there," our knowledge of it is constrained by the kind of beings we are, the kinds of senses our bodies are equipped with, the way our minds work, or the kinds of social practices, language above all, which we have developed in order to interpret the world and to act and survive in it. Human knowledge, this is to say, is not direct, universal, or absolute, but is constrained by human nature and needs or, at the very least, by the constitution of human reason.

Before turning to pragmatism--the third and, it must be said, the favored position--it may be worth recalling why this brief excursus into epistemology was necessary in the first place. For empiricists, no truth about the world can be established other than on the basis of experience. Rationalists, on the other hand, take the view that we may have a priori knowledge of the world. This means, literally, knowledge before experience; but the sense of "before" is not temporal. Rather, the rationalist claim is that there are things we know independently of experience, so that, for example, we might confidently dismiss the evidence of our senses as illusory if it appeared to be supporting a proposition we knew, a priori, to be false. Sometimes we cannot believe our eyes; rationalism helps to justify such disbelief. There is also a stronger implication to be drawn from a rationalist position: there may be a class of a priori truths which are not true simply by virtue of the meanings of their constituent terms, but inasmuch as the world is as they claim it to be. Statements expressing such truths are "synthetic" in the sense that their predicates are not just restatements of their subjects but add something new; they are also a priori, known independently of experience. Synthetic a priori knowledge, if it makes any sense at all, would seem to consist in all those things we know about the world because they must be true of any possible world which, being as we are, we could know about.

It will shortly become clear that Waltz commits himself to positions in the first chapter of TIP which make it quite impossible to regard him an empiricist. At the same time he is at pains to avoid an outright rationalist position, and for good reason. His theory therefore purports to identify and explain an orderly reality beneath the flux of events without making entirely clear either the ontological status of this reality or how we may have knowledge of it. This, coupled with a positivist methodology with empiricist implications, has led some critics to brand him, confusingly, as an empiricist. Spegele, for example, accuses Waltz of espousing an empiricist theory of meaning while in the same paper quoting methodological declarations from his work that are more consistent with a pragmatist, or even a rationalist epistemology (Spegele 1987: 196, 201).

All this might seem mere nit-picking. Yet how these questions are decided makes a tremendous practical difference to the manner in which one may properly set about the study of international relations, to the status and implications of truth claims, and to views of the relation between investigation and policy. Accepting the empiricist position, one may set out to hypothesize and to test one's beliefs against the observed world; accepting a rationalist viewpoint one is more likely to place emphasis on social order as the working out of rule-governed behavior in a world of happenstance and contingency. The first position aspires to greater objectivity, and hence to a different relationship with policymakers, than the second.

The distinction is comparable to that between the two routes leading to knowledge of natural law as outlined by Aquinas. The first lay through observation of the extent to which different peoples shared common behavior and institutions; the second, through a quasi-deductive process of deliberation from allegedly self-evident fundamental principles (Aquinas [1973]). The results of the first process may effectively be delivered by advisers as an intermediate intellectual good to statesmen who have not themselves done the work. The second requires that statesmen themselves participate in the deliberative process rather than simply act as its executants. This is because in the Thomist scheme of things the contingency of human affairs means that arguments from first principles of natural law, unlike theoretical deductions, are indeterminate in outcome. It matters just who does the deliberating, and good statesmen are those whose deliberations are most prudent. Waltz shows great reluctance to decide between the two paths, his world leaders seem much more molded by than molders of the structures in which they operate, and the methodological compromises stemming from this epistemological fudge pervade the whole of the TIP. But they are perhaps more than commonly evident and most easily exposed in the analogy between the balance of power and microeconomics, shortly to be considered.

This digression has delayed consideration of the third epistemological viewpoint, pragmatism, which, alongside empiricism and rationalism, has dominated philosophical debate. It is time to get back on course. Pragmatism is in some respects a compromise position, in others a rejection--some would say an evasion--of the traditional question concerning the grounding of knowledge. It is less concerned with the source of knowledge, whether in experience or in human rationality, than with the consistency of one knowledge claim with another. Rationality is founded on certain basic logical rules, so the argument runs. One such is noncontradiction. We cannot simultaneously hold it to be the case that both "a" and "not-a," where "a" is some claim about the world. The dialogue between experience and reason continually poses questions about which of our beliefs may be preserved and which must be rejected, but it does not absolutely determine the answers. We are left with choices; and our criteria for choice always include the question: "If we drop this or that cherished belief--if, for example, we drop the claim that the earth is the center of the solar system and replace it with a Copernican claim placing the sun at the center--just what do we gain and what do we lose? Which of our old beliefs will have to go because they are incompatible with the new claims? What previously anomalous observations from experience can now be made sense of systematically?"

Why should there be anomalies? Why should it ever be the case that one knowledge claim is incompatible with another? If empiricism were correct, incompatibility could arise only from error. It should therefore be resolved, not by arbitrary choice, but by scientific investigation, which might consist in procedures of hypothesis and testing of theory or else the cleaning up of the language of investigation so as to yield a pure descriptive language in which the whole set of truths about the world could, in principle, be rendered in a set of mutually consistent propositions.

It was precisely this logical positivist project of developing a logically perfect language, perfect, that is, in its correspondence to and total representation of the world, that foundered between Ludwig Wittgenstein's Tractatus Logico-Philosophicus of 1921 and his later Philosophical Investigations (though it must be said that the worm was already in the bud in the earlier work, as Bertrand Russell observed in his prefatory comments on Wittgenstein's distinction between meaningful propositions and the inexpressible expressiveness of language) (Wittgenstein 1961:xxi). By his later years, Wittgenstein had come to the view that certainty did not stem from the faithful execution of a sequence of logical steps or from careful observation but from a belief's being "anchored in all my questions and answers, so anchored that I cannot touch it" (Wittgenstein 1969:16).

If clear and transparent expression of knowledge about the world is not to be had, whether because of the fallibility of our senses, the nature of our rationality, or the shortcomings of language, then epistemological emphasis shifts from correspondence to coherence, the methodological emphasis from testing to action. The pragmatist question emerges: are our beliefs consistent one with another, at least to the extent that we may have mutual understanding and some degree of social order? Empiricism and rationalism survive in the background, one may say, so long as pragmatism strives for a bundle of propositions which is to be considered maximally consistent with experience, or rationality, or both. This consistency might seem to let empiricism or rationalism back in, but perhaps this is only so if consistency is read in terms of some strictly spatial or geometric metaphor involving the point-to-point projection or mapping of propositions constituting a belief system on to the world around us or out of our minds. If consistency is modeled instead on the image of game playing, defined by the ability to make successive, never pre-determined steps within a set of rules, then this problem may seem less important.

Later it will be argued that pragmatists are no better off in any attempt to use hypothesis or analogy to test a theory than are empiricists, precisely because they cannot justify the projection of theory on to the world through hypothesis any better than empiricists can. But this is not nearly such bad news for pragmatists as for empiricists, since the former group can still fall back on the consistency of belief with action, thereby dodging ontological and epistemological wrangling. "Children do not learn that books exist, that armchairs exist, etc., etc.,--they learn to fetch books, sit in armchairs, etc. etc." (Wittgenstein 1969:62). Waltz's epistemological error, in a nutshell, is not to have disentangled his pragmatism from remnants of rival positions; his consequent methodological error, not to have realized the flexibility conferred on political discussion by a thoroughgoing pragmatism and the impediment, by contrast, presented by a positivist approach to theory and testing.

Going further, there are advantages in moving from a traditional pragmatist position, still within the traditional epistemologically oriented philosophical debate, to a more ontologically oriented view which locates the interaction of material and conceptual entities in a broadly defined sphere labeled "action," which subsumes language. This removes stress from epistemology and places it instead on ontology: are the material instruments and constraints encountered by any human design the same kinds of things as social institutions, language and personality included? If not, then how do they interact? It may have to be conceded that the dualism we hoped to bury alongside the epistemological project (the search for certain knowledge) crops up again when attention is moved to ontology. Provisionally, the changed emphasis might nevertheless be defended because it facilitates treatment of language as a medium both of reflection and deliberation, theory and practice, knowledge and action.

 

Waltz on Theory

The reason for these philosophical preliminaries will now be spelled out through an examination of Waltz's views about the nature of theory, which, in turn, will help explain his use of and expectations from the device of analogy. The strategy adopted against Waltz might, in the terms of the preceding section of this chapter, almost be characterized as pragmatist; its central thrust is not that Waltz was wrong, but that he tried to construct a single picture using the pieces of not one but two jigsaws, and that this led to needless confusion.

An important way of expressing what is more immediately at stake in all this is to speak of the degree of privilege accorded to different classes of cause. The extremely sparse structural theory of international politics Waltz developed does not suggest that structural forces are the only causes influencing world politics. That would be absurd. Instead, an order of causation is established. The structure of international politics is said to set up constraints within national political systems which influence the choice of leader in great powers and govern the behavior of each leader while in office. The reader is asked to take comfort in the thought that "those who direct the activities of great states are by no means free agents. . . The pressures of a bipolar world strongly encourage them to act internationally in ways better than their characters may lead one to expect" (Waltz 1979:176). And Waltz very appositely cites Richard Nixon as an example. But this is dangerously close to saying that the security of the structure gives states and world leaders leeway for error and frailty, which cannot be the case if the task of continually reproducing social structures falls upon individuals, as we have contended in Section II. The idea of privilege--of certain classes of cause, of the literal over the metaphorical, of economics over international politics--runs throughout these deliberations and is a principal target of this chapter because in methodology, as in life, privilege exists essentially to resist change and is for this reason in the end not so much wrong as futile.

Returning, now, to Waltz on theory, notice the privilege given to structure over unit, continuity over detail, theory over fact. Notice, too, the privilege given not simply to structural or system-level cause over unit level or individual causes, but also to efficient cause over any notion of final cause or intention. In privileging structure over agent one is necessarily also privileging efficient cause over intention. Structures--whether markets or the balance of power--transform the intentional behavior of uncoordinated actors into orderly but unintended outcomes. Furthermore, structuralist accounts provided by modern economics or by post-Darwinian biology provide powerful reasons for regarding the apparatus of final cause and teleology as redundant when explaining this unintended order.

It is very hard to see how any such privileging could emerge from a radically empiricist account of causation as mere constant conjunction; and indeed, Waltz takes a strongly anti-empiricist line on the nature of theory in the early pages of his book. He goes out of his way to reject naive positions which would represent theory as the mere cumulation of observations of law-like regularities in the world (Waltz 1979:6). Facts do not speak for themselves. In the manner of Kant, whose work he knows well, Waltz outlines two paradoxes before proceeding to a slightly shaky solution (Waltz 1962). Thus Waltz recognizes that "Knowledge. . .must precede theory, and yet knowledge can proceed only from theory" (Waltz 1979:8). He goes on to admit that "facts do not determine theories; more than one theory may fit any set of facts." Conversely, "theories do not explain facts conclusively; we can never be sure that a good theory will not be replaced by a better one" (Waltz, 1979:9).

At this point, all the preliminary moves have been made for Waltz to embark on a discussion of the status of the assumptions of any theory that would place him firmly in one or other of the traditional epistemological camps. Are they analytic or synthetic; are they known a priori or a posteriori; are they necessarily or contingently true (if indeed they can be true at all)?

But this discussion does not take place. Faced with deciding whether the origins of scientific theory lie in representation of the world or in some projection of rationality on to the world, Waltz turns instead to a consideration of scientific creativity. "The longest process of painful trial and error will not lead to the construction of a theory unless at some point a brilliant intuition flashes, a creative idea emerges," Waltz claims. This is the end of the line. He continues: "One cannot say how the intuition comes and how the idea is born" (Waltz 1979:9). But however difficult it may be to deal with the question of the logical and epistemological status of theory it is not a problem to be neglected with impunity, and Waltz's failure to pursue it is unfortunate.

Confusion arises in the following manner. While firmly avoiding the empiricist horn of the Kantean dilemma, Waltz is no more anxious to impale himself on the dogmatist or idealist horn directly. But he makes commitments in his first chapter that lead him well down this second road before seeming to double back in a final subsection dealing with the testing of theories. Here he embraces a positivist methodology which seems to presuppose the empiricism he has just rejected.

Let us examine these commitments. Theories, for Waltz, are made up of two sorts of statements, descriptive and theoretical. Theoretical statements are those which include theoretical terms or notions. These have, Waltz insists, no meaning outside the theory (Waltz 1979:6). Their meaning, it seems, arises from differences within the discourse of the theory rather than from reference to a world outside of the theory. The truth of the statements containing them would appear to be a matter of internal coherence rather than external correspondence. Surely this means that such statements--the axioms of a theory--must either be analytic or, just possibly, synthetic a priori judgments (which is to say truths known independently of experience which must nevertheless necessarily be true of any possible world in which entities of the kinds to which they refer are to be found); but there is nothing in the text to indicate that Waltz intends to take this latter, and difficult, route, while there is much that points in the alternative direction.

For example, Waltz maintains that theories are not the kinds of entities that can be judged true or false. Rather, they are successful or unsuccessful, to be judged by "the number of previously disparate empirical generalizations and laws that could be subsumed in one explanatory system" and "the number and range of new hypotheses generated" (Waltz 1979:6). This begins to sound very much like a form of Quinean pragmatism in which truth figures first and foremost as coherence, constrained by experience and by some notion of the marginal cost of holding the very last view that can be tacked on to the hem of a theory before it starts to drag in the mud.

Any reference which theoretical statements make to a possible world is, then, not a matter of one-to-one reference of proposition to fact or mirror-like reflection of reality in theory, in which truth arises unproblematically out of being, but rather an active, shaping, interpretative, or, in Waltz's terminology, an explanatory relation between human reason and the world. "The reason why the use of the expression 'true or false' has something misleading about it--Wittgenstein once remarked--is that it is like saying 'it tallies with the facts or it doesn't,' and the very thing that is in question is what 'tallying' is here" (Wittgenstein 1969:27).

(Note how much more active a part in scientific activity this account of theory gives to scientists, as compared with their rather passive role in empiricist accounts. Scientists are the judges of what is explanatory. Scientists are, it seems, above the game. Scientists--and not the world--give meaning to terms; but in Waltz's account this is done, to his evident disappointment, on the basis of "brilliant intuition" defying further explanation [Waltz 1979:9].)

Where is Waltz's grounding of theoretical discourse in the subjectivity of the scientist to lead us? Perhaps to a sociological account, in the manner of E. H. Carr, of the origins in history and group interest of some of these "brilliant intuitions," plunging scientists back into the objective social world in which they act and from which they derive their intuitions? (Carr 1939). Perhaps to some variety of scientific realism in which the reference of scientific terms is rooted in the continually evolving research community? (Hollis and Nell 1975; Bhaskar 1979). Perhaps into an argument for a broadly dialectical social science construed as conversation, edifying philosophy, or deliberation? (Collingwood 1939; Rorty 1980; Aquinas 1973).

But it turns out to lead to none of these things. Waltz takes cover behind one of the least attractive stratagems of pragmatism, intuitionism. Creativity is to be regarded as beyond explanation. There follows an abrupt U-turn, as theory is welded back to the external world through a positivist characterization of the testing of hypotheses (Waltz 1979:13-16).

 

Waltz on Testing

Since no theory can be directly tested, so the argument runs, it is necessary to test hypotheses derived from the theory. Successful hypotheses strengthen or confirm a theory, though they can never render it invulnerable. Failures may lead to modifications--dare one say improvements?--but may, on the other hand, accumulate to the point where an alternative paradigm suggests itself.

But Waltz is evasive about the relation of theory to hypothesis. If hypotheses are to be testable against experience, then at a minimum they must be synthetic. This is to say the truth or falsehood of the claims they make must be such as cannot be determined wholly by conceptual analysis. The claim that all bachelors are unmarried is simply not worth testing. If Waltz were the positivist that he has sometimes been taken to be, then he would face the problem of how to derive synthetic hypotheses from the analytical (and therefore empirically vacuous) axioms of his theory.

An obvious stratagem to adopt here would be to suggest that this could be achieved through the reference of terms defined within the axiom system to phenomena in the world. But any such reference would appear to require criterial statements to help identify states, wars, alliances, and other entities central to the theory unambiguously, and these in their turn would have to be either synthetic or analytic. And if such statements are analytic, Hollis and Nell insist, "they do not serve to anchor terms to the world; if synthetic, they do not state the sort of criteria required" (Hollis and Nell 1975:96). Making a closely related point, Wittgenstein once remarked that "the same proposition may get treated at one time as something to test by experience, and at another as a rule of testing" (Wittgenstein 1969:15).

There are two routes out of this predicament. The first requires acceptance of the radical contingency of language. Naming is exposed as arbitrary. There may well be given divisions and patterns in both the natural and the social worlds, but the job of language is not simply to approximate to these, nor can approximation to them ever be simple, because they may not all be captured at one and the same time and, to some extent, they are changed by the manner in which they are named. The mere naming of things is already an exercise of will.

The second view is a form of scientific realism in which science, including social science, can be grounded through real definitions and axioms that amount to truth claims about possible worlds. The gist of this second position may be summarized here as the suggestion that the potential manipulative power of axiomatic systems is such that it is generally worth trying to discover whether the world we find ourselves in is one of the possible worlds for which a given theory happens to yield plausible interpretations or a reliable guide to action. In practice, of course, one never goes out cold to try out a theory and its attendant descriptive language on the world. There is instead a gradual exploration of fit in which behavior, including linguistic behavior, and the world of efficient cause and physical systems accommodate to one another. The great difficulty with such a position, signaled earlier in this chapter, is that it nicely skirts old epistemological wrangles only to land fair and square in analogous and equally intractable problems of ontology and reference: what kinds of things (material? conventional?) are real, and are they all similarly real? However purposeful our language may be, and however infused with metaphor, dead and alive, do some at least of its descriptive terms mark natural kinds of real entity--the chemical elements, perhaps? Finally, must that seemingly privileged reference of descriptive terms to reality be limited to material reality, or might it extend to the kinds of real-though-immaterial entities (states, persons, and the like) that could provide the foundations of social theory?

The positivist project of testing theories against the world poses the question of the status of descriptive language and the way language refers to or latches on to the world in an especially stark way by placing the whole of this burden of accommodation on language, which is expected faithfully to reflect the reality of the external world, and claiming language as being primarily or basically concerned to carry out this charge. Waltz appreciates this and makes it quite clear that he does not wish to travel the empiricist route. He accepts that we experience the world only through language which is already theoretical. "Even descriptive terms acquire different meanings as theories change," he suggests, and "changes of theory produce changes in the meaning of terms, both theoretical and factual ones" (Waltz 1979:12). Neorealist theory, it seems, is happily adrift from the world. In his methodology, as in his epistemology, Waltz turns out to be a transitional figure, caught like Milton Friedman or Karl Popper, taking shelter under the umbrella of American pragmatism (McCloskey 1986:10; Cottingham 1984:141-43). This has been, in part, the specific predicament of a long generation for whom the experience of totalitarianism had made rationalism uninhabitable. The same generation produced, not by coincidence, classical Realism and structuralist social science.

 

Economics as "More Developed Theory"

The reason for this detailed examination of Waltz on theory and testing has principally been to expose the inconsistency of his positivist approach to testing with the loosely pragmatist account of knowledge and of the nature of theory which he provides and with which we sympathize. It is no good looking to Waltz himself to provide clear criteria for "more developed theory" which can then be applied to modern microeconomics in order to see whether it is, as he hopes, up to the job of acting as senior partner in the analogy. Waltz manages to make pragmatism seem a compromise between empiricism and rationalism stumbled across in the dark, whereas it is a far more viable and autonomous position than this description suggests.

But it is worse than this. First of all, the incoherence of his own position makes it difficult for one to identify some significant group of economists of a methodological disposition to which Waltz could clearly assent. Secondly, there is in any case no general agreement about the status of theory among economists. Moreover, such agreement as there is turns out to suffer from much the same schizophrenic oscillations between empiricism, rationalism, and pragmatism as does the position adopted by Waltz. It is under attack. The interesting thing--at this point one hardly knows whether it is encouraging or discouraging--is that most economists (and a good few political scientists) still seem quite unperturbed by the rising tide of methodological uncertainty, even though global ideological warming suggests that intellectual defenses everywhere are in for prolonged battering from the waves of inquiry. "It will not do to say about the methodological rules of economists, as a professional philosopher might, 'No one believes that stuff anymore.'|" (McCloskey 1986:11). Does this indicate that they are simply obtuse, getting on with the job, made secure by their privileged status as advisers to powerful organizations? Does it indicate rather that the profession is shifting gradually to the adoption of a new methodological paradigm, a new way of accounting for and justifying the procedures customarily adopted in the course of practical work?

To most outsiders, economics appears much more self-confident and unconcerned with methodological wrangling than other social sciences. They have often been dismissively characterized en bloc as empiricists. "Economists," as Michael Nicholson so charmingly puts it, "tend to be more cheerfully empiricist than political scientists" (Nicholson 1985). Hollis and Nell, the latter no outsider, cast the whole Neoclassical tradition into what they regard as the pit of empiricist epistemology and consequential positivist methodology (Hollis and Nell 1975).

Yet this is gross oversimplification. Economists differ widely on the importance and logical status they attach to empirical testing of theory. Many are out and out rationalists with no more than a passing interest in whether this world happens to be one in which their theories apply. Yet those in the profession most closely concerned with advising firms and governments have on the whole have been inclined to be more pragmatic than the rhetoric of their textbooks and methodological debates would indicate. Models, even where they are not open to falsification in principle through the testing of hypotheses, tend to be heavily protected against the cold, just in case. There is an abiding rationalist concern with the internal consistency and elegance of models. The position is a curious one, summarized from one perspective by Marc Blaug, who has claimed that "for the most part, the battle for falsificationism has been won in modern economics [so that the] problem now is to persuade economists to take falsificationism seriously," and from a quite opposed perspective by Donald McCloskey, who notes with approval that "falsification, near enough, has been falsified" and goes on to remark upon the pragmatist tendencies indicative of an attempt to escape the bounds of positivism in Milton Friedman's celebrated 1953 essay, "suggesting that modernism cannot survive intelligent discussion even by its best advocates" (Blaug 1980:260; McCloskey 1986:10, 15).

How can Waltz seriously appeal to, as "well developed," an inquiry which is just as riven by methodological dispute, and of the very same kind, as his own field? Any claim that one area of inquiry is more developed than another clearly requires criteria for judging scientific progress, some sort of meta-methodology; but Waltz does not provide this and one wonders if it is to be had.

 

States and Firms

It now remains only to expose two more problems involved in Waltz's use of the analogy with microeconomics before moving on in the next chapter to consider an alternative way of regarding the relationship and sketch the kinds of microeconomics this might admit.

The first of the two methodological problems concerns the supposed similarity of form between balance of power theory and microeconomics. "It is likeness of form--Waltz claims--that permits applying theories and concepts across disciplines" (Waltz 1979:55). But one cannot at one and the same time hold on to the notion that states are the major units in the international system and the claim that the balance of power and the market are similarly structured. If the first of these claims is true then economics is in a very important sense subordinate to politics because states set the terms on which markets operate. Politics, insofar as it consists in the behavior of states, may be held constant, temporarily frozen, to facilitate economic analysis, but because they are in this sense dependent institutions, markets must be regarded as only contingently anarchic. The balance of power by contrast is necessarily anarchic. While states may impose their authority upon firms, no institution has equivalent authority over states.

If, on the other hand, the market is to be regarded--like the balance of power--as necessarily anarchic, then it can only be because property rights, and from them the whole pattern of production and exchange, are to be seen as pre- or extra-political social institutions, able to function, in principle, in a world without states. At this point, theories of the need for a hegemonic power to underpin the liberal world economy are threatened, and it is open to thoroughgoing economic liberals to argue that the state system is merely one, and perhaps not the best, possible way of organizing authority relations globally. They might easily argue that the sovereign state is a quite distinctive and recent invention, certainly a sight more recent than systematic exchange of goods between otherwise autonomous societies.

It may be helpful at this point to run through two arguments in a little more detail, the first of which, called "Hobbes," regards markets as contingently anarchic, while the second, called "Rousseau," regards them as necessarily anarchic. In distinguishing these two positions I shall hope to unpack the seeming paradox with which Douglass North opens his chapter on the neoclassical theory of the state. He writes: "The existence of a state is essential for economic growth [Hobbes]; the state, however, is the source of man-made economic decline [Rousseau] (North 1981:20). The interpolations in square brackets are mine; they are, of course, no more than nick-names.

Both views start from the question of what makes it possible for markets to function. One reply is that without the state to allocate and protect property rights, guarantee the enforcement of contracts, and provide standardized weights, measures, and money, markets simply cannot work. It is hard to imagine voluntary exchange in a Hobbesian state of nature where life and property are radically insecure. It is on this basis that Frederick Lane and, later, Douglass North, developed the idea of the state as a special kind of firm, the essential functions of which were to deploy coercion in such a way as to create zones of security and set the permitted levels of monopoly in a whole variety of subordinate economic sectors within such zones (Lane 1966; North 1981:20-32). Waltz appears to share this view. "When the crunch comes," he writes, "states remake the conditions by which other actors operate" (Waltz, 1979:94).

These might seem perverse examples. Are not Lane and North to be regarded as imperialist economists, arch-enemies of political Realism, for their attempt to cast states as firms and, implicitly, to explain political phenomena by applying economic rationality? Are they not reductionists by the useful criteria set up by Waltz? Perhaps so. Yet the key feature of political Realism which is retained, even in their economistic vision of world history, is the privilege given to the state. The state, albeit an economized state, is still required as the actor without which the market system could not operate. Good structuralists that they are (in a general way, if not by the specific criteria set by Waltz), Lane and North respond to the apparent incompleteness of the world normally analyzed by economists by bringing the state and coercion within economics and subjecting them to the kind of analysis paradigmatically applied to the production and exchange of private goods. The market for protection from violence (or the market for public goods such as deterrence or equity) may be a funny sort of market, but that is surely better than no market at all. This strategy promises to restore the autonomy of economics and extend its scope at one and the same time by releasing it from dependence on an exogenous initiating force. It is Realist by its state-centricity and structuralist by its reliance on and privileging of microeconomic analysis.

A second possibility, referred to here as "Rousseau," but available in paperback as "Paine," is to argue for the possible emergence of voluntary exchange on the basis of mutual interest or custom in a pre-political civil society. This argument has a very lengthy pedigree. Unlike "Hobbes," it offers an economic system that, however much it may interact in practice with the political, is every inch as anarchic in essence and origin as is the world of states. But it does so at a price. It requires the consistent Realist to take a step back from customary assertions of the privileged position of the state in international relations and from the more general priority assigned to political power over forms of social power based in other social institutions: in gender and seniority (patriarchy), in language (rhetoric), or in markets (property).

In sum, by employing the argument by analogy between microeconomics and the balance of power in the way that he does Waltz is driven toward an uncomfortable but perhaps unnecessary choice between two Realist dogmas: the centrality of state power and the organizing principle of anarchy.

It has seldom been clear in Realist accounts whether the continuing emphasis on international politics as a distinct domain or field of studies is a separatist or an imperialist ploy. If "Hobbes" is right and state supremacy makes politics a kind of master-social-science, then the analogy with microeconomics is spurious because of the contingent character of the anarchy displayed by markets. (Any superiority of economics as "developed science" might well arise only from the limited extent of its scope and therefore be non-transferable.) But Waltz's insistence on the market analogy and a pervading recognition of the variety of sources and expressions of social power in the broader Realist tradition suggest that something closer to "Rousseau" may offer space for a reformulation of the economic analogy, with its undoubted heuristic value, while retaining an emphasis on the distinctive, rather than overweening, nature of state power. The trade-off directs attention toward an issue that has already received attention in Section I, and to which this Section turns in its final chapter, namely the aggregation of power. Our argument there proceeds from the observation that the multi-sectoral approach of Structural Realism resonates with disaggregated state power, argued for in Section I, since states exercise each of their distinctively distributed capabilities not only in their relations with one another but also in their relations with non-state actors whose core activities may be economic, cultural, or coercive.

 

Analogy as Confirmation

The second methodological point referred to at the start of the previous subsection also relates to the permissibility of the economic analogy. It has to do with Waltz's view of the role that reasoning by analogy can play in the confirmation of theory. Confirmation may, on his view, be achieved in two ways. One is by the formulation and testing of hypotheses, and attention has already been drawn to problems in that procedure. But Waltz goes on to argue that "structural theories. . .gain plausibility if similarities of behavior are observed across realms that are different in substance but similar in structure. . . This special advantage is won: International-political theory gains credibility from the confirmation of certain theories in economics, anthropology, and other such non-political fields" (Waltz 1979:123).

There is a certain amount of truth in this. Formally similar axiomatic systems might surely be developed in tandem. But Waltz, if he once admits the problem of getting any grip on the world for purposes of testing theories by the stratagem of developing (synthetic) hypotheses from (analytic) axioms, seems to have only three routes open to him. One would be to admit that any theory is simply a set of analytical truths and is empirically vacuous. Structural similarity between theories then in turn disappears into empirical vacuity. It may persuade people, although it is hard to see why; it certainly confirms nothing. The second alternative might be to seek to ground theory in the world through the rationalist device of essential or real definitions. A variant, rather closer to our own pragmatist position, would be to call an orderly retreat in the face of the difficulties confronting any such attempt to ground knowledge, opting instead for a causal account of reference. This would accept a closer relation between different bodies of discourse and the discrete areas of social practice of which they form a part than prevails between the several and distinct discourses which, traditionally, are regarded as forming a single coherent natural language (Schwartz 1977).

If this pluralist strategy is adopted, and if structural similarities then emerge between one social science and another, it will certainly be a matter of interest and may well change one's view of both fields, but it is hard to see how it will confirm either theory. In the first case, confirmation would still make sense, but only within bounded areas of social action. In the second case, the very notion of confirmation dissolves. Instead of being seen as a more or less accurate representation of action, discourse--including theoretical or meta-discourse--comes to be regarded as part and parcel of that action. It cannot ground or confirm it.

In any case, the procedures laid down by Waltz for testing a theory, which presumably are to be applied equally when employing analogy in its confirmatory role, are deficient. Describing his procedures for testing a theory, Waltz lays out seven steps. The theory must first be stated. Hypotheses must be inferred from it. They are next subjected to tests, care being taken throughout to employ the definitions and terms of the theory. (This begs the questions of definition, reference, and the bridging from analytic to synthetic, with which we have been concerned in this chapter). Exogenous variables must be eliminated or compensated for, and crucial tests sought. The last step urges the scientist, in the event that a test is not passed, to "ask whether the theory flunks completely, needs repair and restatement, or requires a narrowing of the scope of its explanatory claims" (Waltz 1979:13).

Here is a procedure crying out for criteria. On what basis is one to make these important decisions? It is never entirely spelled out, and to judge by later descriptions of tests it seems that common sense is to be our guide, which is a polite way of conceding that here, as with his account of the intuitive origins of theory, Waltz is content to accept practical limits to reason. To Hollis and Nell, as realists, this feature of pragmatism seems more like an abandonment of reason. They spell out the extent to which it makes a sham of any pragmatist claim to stand at the tribunal of experience. "When experience conflicts with expectation," they insist, "pragmatism always offers the scientist a choice of how to pay the bill. Our objection," they continue, "was that universal choice includes choice of whether to pay the bill at all, and that statements of the 'price' of revisions [i.e. what belief must be sacrificed if another is to be adopted] should, for the sake of consistency, themselves be revisable" (Hollis and Nell 1975:163).

 

Conclusion

The analogy running throughout Waltz's TIP has been used in this chapter as a diagnostic tool to discover some of the breaking points of his formulation of Realism. Many of the contradictions have been hinted at, but not yet fully explored. It has, for example, been claimed that traces of positivist positions on theory and testing in TIP do not sit well with more pragmatist attitudes on epistemology and methodology in the early chapters. It has been suggested that Waltz does not confront the tension between rhetorical purpose and scientific objectivity as squarely as other (less honest) figures in the Realist tradition, such as Carr. It has been suggested that the pervading comparison between the balance of power and oligopolistic markets, with its nostalgic casting of economics as the most objective of social sciences, exposes a central difficulty of Realism concerning the relation between state power and other forms of social power. Presenting contradictions in this way always leaves open the rhetorical strategy, used to such great effect by both E. H. Carr and Hedley Bull, of suddenly pulling a middle way, a splendid compromise, out of the hat. But compromise or reconciliation of apparently contradictory views is not the only way of having cakes and eating them. Indeed, it is often achieved at the level of policy prescription only by leaving epistemological and methodological issues unresolved, storing up problems for the future. Another strategy is to seek ways of effacing the compromises, leaving the formerly contradictory positions intact, still in conflict one with another, perhaps, but at least in a creative and self-conscious conflict. The view taken here has been that many of the apparent contradictions in Waltz stem from inconsistent, inadequately developed, or mistaken epistemological and methodological positions. The remedy, therefore, would appear to be to establish alternative, less vulnerable foundations. No one builds directly on ground known to be treacherous because of old mine workings beneath it. Instead, it is common practice to build on a concrete raft. No builder mistakes a raft for bedrock, yet the artificial may in these circumstances provide a surer ground than the natural. So next we build a raft.